Trans-spatial phase coupling at the lipid boundary: harnessing membrane biophysics for therapeutic condensate regulation

1 Introduction

The formation of highly dynamic, membraneless biomolecular condensates (BCs) via phase separation—commonly referred to as liquid-liquid phase separation (LLPS)—has fundamentally reshaped contemporary principles of cellular organization. The historical perspective of biomolecular condensation as a bulk-phase phenomenon, where classical Flory-Huggins polymer physics drives the random phase approximation and coacervation dynamics, has since evolved. Contemporary biophysics now demands sophisticated paradigms that capture the profound regulatory influences of physical boundary conditions, specifically the complex thermodynamic topology of the lipid bilayer. This interface between membranes and condensates operates as a contiguous thermodynamic continuum where 3D bulk phase transitions and 2D planar transitions reciprocally dictate one another (Kang et al., 2024). Three interconnected paradigms elucidate the physical mechanics governing membrane-associated phase separation through distinct biophysical lenses.

1.1 Interfacial thermodynamics: establishing the elastocapillary foundation

This paradigm addresses the fundamental physics governing the interaction between a 3D fluid condensate and a 2D lipid sheet. BCs do not merely float adjacent to the bilayer; they wet the interface, dictated by spreading coefficients, Neumann’s triangle, and salinity-dependent transitions from dewetted to completely wetted states (Mangiarotti et al., 2023a). The specific wetting geometry is frequently stabilized by targeted chemical networks, such as hydrogen bonding between specific amino acid residues and zwitterionic lipid headgroups (Zeller and Cremer, 2025).

1.2 Reciprocal material remodeling: the effect of scale-bridging mechanics

Extending beyond flat surface assumptions, this paradigm examines the trans-spatial feedback loop where capillarity physically distorts a highly elastic membrane, while the internal structural order of the membrane reciprocally alters the condensate. Experimental metrology, including LAURDAN phasor analysis, demonstrates that the act of wetting directly alters the local hydration and physical state of the lipid bilayer (Mangiarotti et al., 2023b). Furthermore, the degree of internal lipid packing and local cholesterol content acts as a strict thermodynamic gatekeeper, actively regulating whether a condensate can wet the surface or is mechanically excluded (Mangiarotti et al., 2025).

1.3 Non-equilibrium trans-spatial circuitry: entering the kinetic frontier

Within the living cell, the trans-spatial membrane interface functions not as a static boundary, but as an open thermodynamic system adapting to fluctuating energy landscapes—exemplified by pump-driven energy dissipation at the inner mitochondrial membrane. The membrane 2D lipid boundary operates as an active circuit, dictated by the highly localized, lateral proton diffusion networks generated by ATP synthases (Flegel et al., 2026). Under this paradigm, the static wetting geometries (Paradigm 1) and localized lipid packing alterations (Paradigm 2) can function as direct electromechanical dampeners or modulators of these active 2D energy gradients, fundamentally linking macroscopic condensation to localized bioenergetic output.

The modulation of ATP synthase output represents a highly complex therapeutic objective, fundamentally dictated by the capacity to precisely titrate bioenergetic throughput. This Perspective article explores a window of opportunity presented by the biphasic, non-equilibrium trans-spatial regulation of the ATP synthase. By evaluating how membrane-wetting condensates act as elastocapillary resistors within a 2D lipid circuitry, we define a potential empirical roadmap for deploying interfacial linactants to mechanically tune these vital boundary conditions.

2 Trans-spatial membrane coupling in mitochondria

Operating far from equilibrium, the inner mitochondrial membrane functions as a highly confined 2D planar circuit that can accelerate bioenergetic reactions. Contemporary ultrafast metrology has significantly expanded the biophysical boundaries of bioenergetics, revealing that the lipid bilayer operates not merely as an inert dielectric barrier, but as a highly active, spatial participant in energy transduction. Utilizing time-correlated single-photon counting (TCSPC) coupled with membrane-tethered excited-state photoacids (C₁₂-HPTS), researchers have successfully isolated and tracked proton migration dynamics directly at the lipid-water interface. When the reconstituted F_OF₁ ATP synthase is inactive, protons exhibit a fractal lateral diffusion dimensionality (d = 2.7 ± 0.2), reflecting significant proton escape perpendicular to the membrane into the bulk aqueous phase. However, upon activation via a transmembrane pH gradient, the enzyme initiates a highly efficient proton consumption cycle that forcefully collapses this diffusion dimensionality strictly to the two-dimensional interfacial plane (d = 1.8 ± 0.3). This profound 2D confinement establishes a localized gradient that channels protons directly along the membrane surface, achieving an ultrafast initial transfer rate (k _PT = 15.9 ± 2.3 ns^-1) that far exceeds bulk water transport capabilities (Flegel et al., 2026).

Crucially, this localized 2D circuitry is mechanically integrated with the active conformational dynamics of the enzyme itself. Operating as a biological nanorotor with a rotational frequency of approximately 20 Hz, the active ATP synthase drives massive, non-equilibrium height fluctuations at discrete membrane hotspots (Almendro-Vedia et al., 2017). This continuous mechanical pumping action softens the surrounding bilayer and dynamically alters local curvature, promoting the collective synchronization of ATP synthases at highly curved cristae ridges. Essentially, this curvature-protein coupling provides an electrostatically and mechanically tuned interface capable of sustaining rapid, directed 2D proton transport, ensuring operational efficiency despite the massive fractal tortuosity and restricted diffusion parameters imposed by the highly macromolecularly crowded mitochondrial matrix (Bulthuis et al., 2023).

Unlike simplified in vitro models, the inner mitochondrial membrane features a highly specialized, cholesterol-deficient lipid composition uniquely enriched in cardiolipin (CL). Within this specialized and crowded matrix, condensate wetting inevitably dehydrates the interface. The loss of the hydration layer severely restricts the lateral diffusion of membrane lipids and proteins (Chattopadhyay et al., 2021). While CL effectively compensates for desiccation stress by transitioning into inverted hexagonal (H I I) tubes to facilitate continued, rapid proton movement (Rani et al., 2025), the overall restriction of planar macromolecular mobility compromises optimal ATP synthase dynamics and supercomplex assembly (Romanova et al., 2025). Exogenous interfacial linactants including melatonin operate as multivalent plasticizers; by partitioning into the boundary, they rehydrate and fluidize the interface, allowing CL to relax back into a continuous lamellar state that rescues 2D planar lateral diffusion.

2.1 The biphasic effects of membrane dynamics on condensate wetting

The operational efficiency of this active 2D proton circuitry is entirely subordinate to the physical boundary conditions imposed by biomolecular wetting. Macromolecular wetting represents a scale-bridging thermodynamic phenomenon wherein the adhesion energy of a liquid droplet competes directly against the surface tension and bending rigidity of an elastic lipid bilayer (Lipowsky, 2013; Mangiarotti et al., 2023a). As a bulk three-dimensional condensate contacts a membrane, the system balances along a morphologically diverse continuum defined by distinct dewetting, partial wetting, and complete wetting transitions (Mangiarotti et al., 2023a). Classical lipid biophysics dictates that high-curvature geometries, such as the apices of cristae ridges where ATP synthases cluster, are intrinsically fluid and rich in liquid-disordered (L_d) packing defects (Wang et al., 2024). These localized L_d defects do not merely scaffold proteins; they function as profound thermodynamic sinks. This is driven by localized changes to lipid packing defect distributions that minimize total system free energy (Cui et al., 2011), achieving an extraordinary geometric protein sorting capacity that can exceed thousands of percent (Larsen et al., 2020).

For intrinsically disordered proteins like mitochondrial transcription factor A (TFAM), this highly negative curvature can also dictate the spatial boundaries of phase separation. Rather than relying on classical 3D bulk-phase saturation, TFAM undergoes a strict curvature-mediated prewetting transition. The specific geometric topology of the cristae lowers the effective chemical potential and the nucleation energy barrier, actively driving the transition from a dilute adsorbed layer into a tightly packed, high-density condensate (Feric et al., 2021; Hu et al., 2026). In this strongly wetted state, multivalent protein-protein cohesive forces dominate the matrix to safely compact the mitochondrial genome (Aththawala Gedara et al., 2026), physically excluding or kinetically restricting transcriptional machinery from the highly oxidative environment of the active pump (Feric et al., 2022).

Intriguingly, the interaction at this three-phase contact line is governed by a bidirectional feedback loop of reciprocal material remodeling. While BCs exhibit a strict thermodynamic preference for wetting loosely packed L_d domains, the physical act of wetting profoundly deforms the underlying lipid architecture. Because this condensate–membrane interaction is largely mediated by the physical state of water at the interface, highly ordered L_o domains rich in cholesterol and saturated lipids—which are intrinsically packed and dehydrated—exhibit a markedly reduced affinity for phase attachment (Mangiarotti et al., 2025). Mechanistically, dictated by the force balance of Neumann’s triangle, the adhesion free energy of the dense droplet exerts continuous elastocapillary stress on the membrane. This capillary action physically dehydrates the immediate membrane interface. Empirical metrology utilizing spectral phasor analysis of environmental polarity probes (e.g., LAURDAN) confirms that wetted membrane regions experience a sharp decrease in water dipolar relaxation and a corresponding drop in the apparent dielectric constant. This localized dehydration forces the wetted lipid bilayer to transition out of its native, highly fluid L_d state, rapidly compressing the lipids into a tightly packed, L_o or gel-like phase configuration (Mangiarotti et al., 2023b).

2.2 The efficient bidirectional regulation of energy conversion and mtDNA transcription by membrane condensates

The wetting-induced phase inversion is utilized by mitochondria as a highly efficient, calibrated, self-regulating circuit breaker for mitochondrial energy conversion and mtDNA transcription. Under an optimal, energized state, the ATP synthase requires the highly fluid L_d phase to sustain its ultrafast, 2D lateral proton diffusion circuitry and to mechanically dissipate its 20 Hz non-equilibrium height fluctuations. This localized kinetic coupling directly supports the rotating motor, promoting ATP synthesis even when the surrounding bulk gradient is mathematically insufficient (Flegel et al., 2026). Wetting of membranes by 3D BCs including TFAM can act as an elastocapillary resistor at the cristae apex (Feric et al., 2021; Hauer et al., 2026). The physical dehydration of the interface modulates the transition of the fluid lipid leaflets beneath the condensate into a highly packed, rigidified, L_o gel-like state that severely impedes lateral proton mobility. In gel-phase anionic lipid environments, the activation energy for proton transfer plummets to 11–14 meV; this creates an almost activation-less, yet highly trapped, local proton state (Ramanthrikkovil Variyam et al., 2025). Counterintuitively, while an inactive enzyme typically allows protons to escape into the 3D bulk aqueous matrix (d → 3) (Flegel et al., 2026), we propose that the extreme dimensional restriction imposed by the rigidified gel-boundary traps the protons, dropping the diffusion dimensionality toward one (d → 1). This creates a quasi-one-dimensional track that drastically increases the geminate proton recombination rate (k _rec) with local surface defects, stalling the lateral proton supply and effectively choking the bioenergetic output of the rotor without compromising the global bulk gradient.

The completion of this mechano-transcriptional rheostat occurs when localized bioenergetic demand shifts and the inner mitochondrial membrane flattens. As the cristae geometry loses its extreme negative curvature and assumes a planar or convex topology, the lipid packing naturally tightens globally into an L_o-dominant state (Risselada and Marrink, 2008). Because condensates are thermodynamically repelled by tightly packed L_o domains, the TFAM condensate can no longer sustain a tightly wetted interface. The nucleation energy barrier rises precipitously, forcing the condensate to dewet and undergo macro-phase detachment from the bilayer (Mangiarotti et al., 2025). This complete release from the surface boundary conditions permits the compressed nucleoid matrix to relax and disperse into a loose, bulk-phase configuration. The resulting increase in nucleoid porosity physically permits the entry of mitochondrial RNA polymerase, enabling active transcription (Hillen et al., 2017; Feric et al., 2022). Ultimately, this thermodynamic cycle ensures that regions of maximum bioenergetic output naturally coincide with genomic compaction and protection, utilizing the physics of wetting to perfectly synchronize structural membrane relaxation with transcriptional activation.

3 Melatonin: the evolutionarily-conserved therapeutic boundary phase-modifier interfacial linactant

The identification and targeted deployment of linactants that can mechanically tune specific boundary conditions may be the lynchpin for the successful translation of this biphasic, non-equilibrium trans-spatial regulation of ATP synthases by membrane-wetting condensates acting as elastocapillary resistors within a 2D lipid circuitry. We propose that melatonin (N-acetyl-5-methoxytryptamine)—a four billion year-old, evolutionarily-conserved indoleamine—is an ideal amphiphilic interfacial linactant, fully capable of physically fluidizing membranes dehydrated by condensate wetting and mechanically directing condensate dynamics via the modulation of lipid domain morphology (Mei et al., 2020; Loh and Reiter, 2021; Loh and Reiter, 2024).

3.1 The empirical landscape of melatonin-induced membrane reorganization

Empirical evidence from model membrane systems demonstrates that melatonin can drive localized bilayer membrane thinning often coupled with the thermodynamic modulation of lipid domain phase boundaries in a concentration-dependent manner. Small-angle neutron scattering (SANS) and molecular dynamics (MD) simulations demonstrate that at lower molar fractions, melatonin partitions into the hydrophobic-hydrophilic interface where it acts as a surfactant, expanding the lateral area per lipid and lowering line tension to stabilize domain coexistence. Conversely, escalating the concentration fraction triggers a progressive increase in hydrocarbon chain disorder and membrane thinning, ultimately shifting the bilayer from localized nanoscopic stabilization toward macroscale phase separation. Importantly, the specific length of the saturated lipid chains used and the complexity of the lipid mixtures dramatically dictate the baseline lipid packing and the subsequent structural response to melatonin.

Within a quaternary neuronal mimic, the 18-carbon saturated chains of DSPC (1,2-distearoyl-sn-glycero-3-phosphocholine) form highly rigid, tightly packed, ordered domain regions at room temperature. Melatonin selectively partitions into these ordered, sterol-rich domains at a low membrane mole percentage (mol%) of 1–3, where all-atom simulations reveal that its hydrophobic indole ring aligns flat and parallel to the membrane interface to maximize stable hydrogen bonding with the lipid headgroups. However, upon escalating into the 7–15 mol% regime, this orientation undergoes a significant shift; the indole ring’s tilt angle broadens, forcing the molecule to insert deeply into the hydrophobic core while its dynamic hydrophilic groups remain engaged with bulk water. Radial pair distribution functions ( g r) confirm that this deep insertion triggers a non-specific lateral redistribution of the indoleamine across both ordered and disordered phases. At the macroscale, SANS measurements demonstrate that this molecular redistribution expands the domain correlation length ξ exponentially from 6 nm to over 100 nm, indicating that melatonin pools into interconnected, honeycomb-like networks that scaffold and stabilize the macroscale domain boundaries (Bolmatov et al., 2026).

The mechanical severity of this orientation shift is intrinsically dependent on the hydrophobic thickness and saturation of the host lipid matrix. In unsaturated 18-carbon DOPC systems, the native structural disorder permits melatonin to expand the apparent area per lipid (A L) linearly as a function of concentration, whereas the absolute membrane thinning effect rapidly reaches a physical saturation limit (Kondela et al., 2021). Conversely, in thinner, 14-carbon saturated DMPC bilayers, the linactant effectively fluidizes the hydrocarbon chains, but the degree of global membrane thinning exhibits minimal concentration dependence even when supersaturated up to 29 mol% (Ivankov et al., 2023). Collectively, these disparate morphological responses confirm that linactants, including melatonin, may not exert a uniform biological signal; rather, they can induce highly variable micromechanical stress dictated entirely by the localized lipid packing density and the precisely partitioned stoichiometric fraction.

3.2 Thermodynamic calibration: the stoichiometric conversion

To map structural alterations to functional bioenergetic outcomes, the in vitro aqueous dose must be mathematically converted to the localized stoichiometric mol% utilizing K p ≈ 2000 as the baseline lipid-water partition coefficient for melatonin (Costa et al., 1995), though partitioning dynamically shifts as wetting alters localized lipid packing. In a closed assay system, the true molar concentration of the modifier partitioned into the lipid bilayer ([M]_lipid) is governed by the total aqueous concentration ([M]_total) and the lipid volume fraction (Φ L):

M l i p i d = K p M t o t a l 1 + K p Φ L

This internal molarity is then converted to a stoichiometric ratio relative to the host lipid’s intrinsic molarity (C L, typically 1.3–1.8 M) (Marsh, 2013):

m o l % = M l i p i d M l i p i d + C L × 100

This calibration yields a continuum highlighting melatonin’s dose-dependent structural effects (Table 1). Pharmacological regimes (<0.1 mol%, ∼1 µM), which encompass baseline pineal-derived circulating plasma levels, restrict activity to classical MT1/MT2 signaling (Bolmatov et al., 2022; Bolmatov et al., 2026). Sub-critical loading (2–5 mol%) initiates mild thinning (0.5–1.0 Å) and 1–2% area expansion without macroscale separation. Moderate loading (5–8 mol%) deepens thinning (1.0–1.5 Å) and expansion (3–5%), stabilizing nanoscopic domains up to ∼20 nm (Kondela et al., 2021; Bolmatov et al., 2026). At 10–15 mol%, ξ grows exponentially past 100 nm into macroscale separation as melatonin forms honeycomb scaffolds counteracting cholesterol’s condensing effects (Kondela et al., 2021; Bolmatov et al., 2026). Progressing into the 20–25 mol% regime, melatonin stabilizes coexisting L_o and L_d phases and introduces direct conformational disorder (Bolmatov et al., 2019; Kondela et al., 2021). As the concentration reaches 27–35 mol%, melatonin eliminates out-of-phase lipid displacements, drastically increasing local raft sound velocity (∼2128 m/s) and establishing a dynamic barrier that drops lateral self-diffusion (D l to 1.26 × 10 ‐ 12 m²/s (Bolmatov et al., 2020). This concentration, inducing transversal thinning up to 7.3 Å, represents the membrane’s maximum geometric limit of conformational disorder while maintaining cohesive structure (Kondela et al., 2021). Progressing past the 29 mol% stability threshold into saturation (>40 mol%) and supersaturation (>60 mol%) in thermodynamic models induces core disorder, driving micellization and ultimate bilayer failure (Kondela et al., 2021; Ivankov et al., 2023).

TABLE 1. Stoichiometric conversion of aqueous melatonin to local membrane concentration.

The estimation of membrane mol% assumes a standardized in vitro cellular lipid volume fraction, recognizing that precise mol% scales strictly with the total lipid mass available in the closed assay system.

Importantly, while mitochondria possess the capacity for in situ melatonin synthesis to natively protect against acute stress (Suofu et al., 2017), these endogenous reserves may be insufficient to function as bulk interfacial linactants under pathological conditions. Therefore, the mechanical domain fluidization and phase boundary modulation described herein represent a distinctly therapeutic modality, requiring targeted, exogenous dosing to cross the necessary stoichiometric thresholds (2–15 mol%) to mechanically resolve pathological condensate wetting.

4 Conclusions and future perspectives

The thermodynamic regulation of BCs via the 2D lipid boundary constitutes the ultimate frontier in cellular organization and bioenergetics. While continuous phase-field models establish that wetted condensates act as rigidifying elastocapillary resistors, interfacial linactants offer a targeted, mechanical intervention. By thermodynamically modulating the lipid matrix, these boundary modifiers physically lower line tension, alter adhesion free energy, and can force pathological assemblies to dewet, thereby restoring localized planar fluidity. However, the therapeutic deployment of any interfacial linactant is intrinsically dependent on a strict metrological understanding of its localized stoichiometric concentration. The capacity of a modifier to alter phase boundaries is a colligative physical mechanism; it is dictated entirely by the partitioned stoichiometric ratio relative to the host lipids.

To ensure metrological rigor and therapeutic fidelity, the mathematical conversion of bulk aqueous dose into a localized mol% should be a primary consideration in vitro explorations. However, the baseline lipidome complexity—including specific concentrations of structural lipids and cholesterol—diverges significantly between rodents and humans. The profound interspecies differences in the inherent bending rigidity and elastocapillary tension of their respective membranes, if not directly addressed via precise stoichiometric calibrations, prevent successful translation of interspecies dosages. Additionally, the accurate replication of ATP synthase membrane dynamics may require cholesterol-deficient, cardiolipin-dense matrices. Without the ability to map the optimal surfactant mol% that stabilizes phase boundaries against the supersaturating mol% that induces severe membrane thinning—ultimately causing significant loss of lipid packing and corresponding bilayer failure—the true metrics driving the mechanics of linactant behavior that enable the therapeutic choreography of the exquisite pas de deux between 3D condensates and the 2D planar boundary will remain obscure.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. The authors declare that generative AI was used during the preparation of this manuscript. The authors employed Gemini 3.1 Pro (Google), an AI-assisted tool to improve language and readability. All content has been reviewed and edited by the authors who will assume full responsibility for the accuracy and integrity of the published work.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Diana Mitrea, DeMix Solutions, LLC, United States

Reviewed by: Diego S. Vazquez, National Scientific and Technical Research Council (CONICET), Argentina